Green emotion - Data collection task STATISTICAL ANALYSIS DATA PREPROCESS Automatic data format check Data quality assessment Data cleaning and filtering Merge data into production database DATA ANALYSIS Descriptive statistics Visual representation Summary tables Advanced Data mining WORK PACKAGE 1 Static Data RAW DATA Structural Data Items: Charging points Electric vehicles Users Dynamic Data Page 1 November 2013
Green emotion Demonstration Regions 2011 2012 2013 Monitored % Electric vehicles 235 536 446 85% Charging points 598 1728 1197 66% Users 269 924 1169 60% Dublin Cork Strasbourg Guipúzcoa Madrid Barcelona Malaga Copenhagen Berlin Karlsruhe/Stuttgart Budapest Pisa Rome Kozani Total uses Charging points 143 704 Electric vehicles Trips 103 983 Charging events 68 327 Each demo region is expanding the network of monitored assets through deploying larger EV fleets and the ongoing installation of smarter infrastructure. As the project matures, the amount and quality of the information improves. 13 Demo Regions from 7 European countries Page 2 November 2013
Vehicles, owners and use Transporter 6% Bus 6% Type of vehicle Vehicle use Motorbike 8% Car 80% Captive fleet 34% Business use 46% Private use 10% Renting 10% Make Model Count Percentage (%) Renault Fluence ZE 227 44,95% Daimler Smart 87 17,23% Think City 27 5,35% Mitsubishi i-miev 24 4,75% Peugeot ion 23 4,55% Tecnobus, S.p.A. Gulliver 520 20 3,96% Vectrix VX1 20 3,96% Piaggio Porter 14 2,77% Castrosua Tempus GNC 13 2,57% Renault Twizy 12 2,38% Type of user Fleet 36% Rent 13% Owner 51% Page 3 November 2013
Charge event characterization Start charging processes percentage 73.32% 26.68% Average charge consumption (IC95%(µ)) Average charge duration (IC95%(µ)) Public access 60.15% installed 26.12% uses Household 10.05% installed 34.37% uses Office parking 25.27% installed 39.51% uses E F Average EV SOC when start charging from 63% to 65% IC95%(µ) Page 4 November 2013
Pattern sequence analysis Represent car life trajectories as sequences of charging, trip and parking events. EV CODE Initial Timestamp Final Timestamp Type... DR1_EV0001 2012-05-12 07:08:49 2012-05-12 07:22:09 trip DR1_EV0001 2012-05-12 07:22:09 2012-05-12 12:45:58 parking DR1_EV0001 2012-05-12 12:45:58 2012-05-12 13:18:36 trip DR1_EV0001 2012-05-12 13:18:36 2012-05-12 13:18:43 parking DR1_EV0001 2012-05-12 13:18:43 2012-05-12 14:00:27 charge... Page 5 November 2013
Charge typology Low Charge (37% Charge events) 83.6% Initial SOC 11.1% SOC increment Medium Charge (38% charge events) 62.2% Initial SOC 26.5% SOC increment High Charge (25% charge events) 39.5% Initial SOC 55.7% SOC increment Based on over 7000 observations Page 6 November 2013
Trip typology Short-Slow Trip (37% trip events) 3.1km trip distance 15.3km/h trip average speed Average Trip (48% trip events) 5.4km trip distance 29.3km/h trip average speed Long-Fast Trip (15% trip events) 28.2km trip distance 62.2km/h trip average speed Based on over 25000 observations Page 7 November 2013 7
Day state distribution Page 8 November 2013 8
Day state clustering 58% 30% 4,5% 7,5% Page 9 November 2013 * Low activity day, few charge and trip events * Trips in the morning, charges in the afternoon * Mainly cars destined to business use and owned by private company. * Charge and trip alternation during daylight and long charges late in the evening. * Trips tend to be short and slow, charges medium * Mainly cars destined to captive fleet use and owned by municipality. * High charge and trip activity. Charges concentrated at night and trips distributed all day long. * Charges tend to be heavy, trips long and fast. * Mainly cars destined to renting use and owned by municipalities. * Trip in the morning, charges during the afternoon with almost no activity at night * Charges tend to be heavy, trips long and fast * Mainly cars destined to business use and owned by private company.
Conclusions and applications The knowledge extracted can be applied: To simulate the user car behavior required in other algorithms such as microgrid optimization. To provide accurate information about the charge cycles in order to estimate the EV battery life span. To realise client segmentation for car manufacturers and electric generation and distribution companies. To help policy makers to regulate and promote the use of EV with objective data. Page 10 November 2013
EVS27 Green emotion Project Session Page 11 November 2013